MCPcopy
hub / github.com/deepspeedai/DeepSpeedExamples / gelu

Function gelu

bing_bert/pytorch_pretrained_bert/modeling.py:51–59  ·  view source on GitHub ↗

Implementation of the gelu activation function. For information: OpenAI GPT's gelu is slightly different (and gives slightly different results): 0.5 * x * (1 + torch.tanh(math.sqrt(2 / math.pi) * (x + 0.044715 * torch.pow(x, 3))))

(x)

Source from the content-addressed store, hash-verified

49WEIGHTS_NAME = 'pytorch_model.bin'
50
51def gelu(x):
52 """Implementation of the gelu activation function.
53 For information: OpenAI GPT's gelu is slightly different (and gives slightly different results):
54 0.5 * x * (1 + torch.tanh(math.sqrt(2 / math.pi) * (x + 0.044715 * torch.pow(x, 3))))
55 """
56 pdtype = x.dtype
57 x=x.float()
58 y = x * 0.5 * (1.0 + torch.erf(x / math.sqrt(2.0)))
59 return y.to(pdtype)
60
61
62def swish(x):

Callers

nothing calls this directly

Calls 1

toMethod · 0.45

Tested by

no test coverage detected